“Have I seen this somewhere before?”
It’s a question teachers have had to ask themselves while marking assignments since time immemorial.
But never mind students trawling through Wikipedia, or perusing SparkNotes for some Great Gatsby analysis, the backend of 2022 saw another challenge emerge for schools: ChatGPT.
The online chatbot, which can generate realistic responses on a whim, took the world by storm by its ability to do everything from solving computer bugs, to helping write a Sky News article about itself.
Last week, concerned about cheating students, America’s largest education department banned it.
New York City‘s teaching authority said while it could offer “quick and easy answers to questions, it does not build critical-thinking and problem-solving skills, which are essential for academic and lifelong success”.
Of course, that’s not going to stop pupils using it at home – but could they really use it as a homework shortcut?
More on Artificial Intelligence
Teachers vs ChatGPT – round one
First up, Sky News asked a secondary school science teacher from Essex, who was not familiar with the bot, to feed ChatGPT a homework question.
Galaxies contain billions of stars. Compare the formation and life cycles of stars with a similar mass to the Sun to stars with a much greater mass than the Sun.
It’s fair to say that ChatGPT let the mask slip almost immediately, as you can see in the images below.
Asking ChatGPT to answer the same question “to secondary school standard” prompted another detailed response.
The teacher’s assessment?
“Well, this is definitely more detailed than any of my students. It does go beyond what you’d expect for GCSE, so I would be very suspicious if someone submitted it. I would assume that they’d copied and pasted from somewhere.”
Teachers vs ChatGPT – round two
Next was a Kent primary school teacher, also unfamiliar with ChatGPT, who gave it a recent homework task.
Research a famous Londoner and write a biography of their lives, including their childhood and their career achievements.
No problem, said ChatGPT, though it’s fair to say that any nine-year-old who submitted the answer below is either being fast-tracked to university or going straight into a lunchtime detention.
“Even just glancing at that, I’d say they copied it straight off the internet,” said the teacher.
“No 11-year-old knows the word tumultuous.”
‘Key decisions’ facing schools
So just as copying straight from a more familiar website is going to set alarm bells ringing for teachers, so too would lifting verbatim from ChatGPT.
But pupils are among the most internet-savvy people around, and ChatGPT’s ability to instantly churn out seemingly textbook-level responses will still need to be monitored, teachers say.
Jane Basnett, director of digital learning at Downe House School in Berkshire, told Sky News the chatbot presented schools with some “key decisions” to make.
“As with all technology, schools have to teach students how to use technology properly,” she said.
“So, with ChatGPT, students need to have the knowledge to know whether the work produced is any good, which is why we need to teach students to be discerning.”
Click to subscribe to the Sky News Daily wherever you get your podcasts
Given its rapid emergence, Ms Basnett is already exploring how her school’s anti-plagiarism systems will cope with auto-generated essays.
But just as teachers must consider teaching students about the benefits and pitfalls of using AI, Ms Basnett said her colleagues should also be open to its potential.
“ChatGPT is incredibly powerful and as a teacher I can see some benefits,” she said.
“For example, I can type in a request to create a series of lessons on a particular grammar point, and it will create a lesson for me. It would take a teacher to analyse the created lesson and amend it, because the suggested lesson, whilst not bad, was not ideal. But, the key elements were there and it could be really useful.
“I could imagine using a created essay from ChatGPT and working through it with my students to examine the merits and faults of the essay.”
Please use Chrome browser for a more accessible video player
2:16
Will this chatbot replace humans?
Dr Peter Van der Putten, assistant professor of AI at Leiden University in the Netherlands, said institutions which chose to prohibit or ignore the technology would only be burying their head in the sand.
“It’s there, just how like Google is there,” said Dr Van der Putten.
“You can write it into your policies for preventing plagiarism, but it’s a reality that the tool exists.
“Sometimes you do need to embrace these things, but be very clear about when you don’t want it to be used.”
‘Bull****er on steroids’
For students and teachers alike, it’s an opportunity to improve their digital literacy.
While it has proved its worth when tasked with being creative, such as to problem-solve or come up with ideas, true comprehension and understanding remains beyond it.
Developer OpenAI acknowledges answers can be “overly verbose” and even “incorrect or nonsensical”, despite sounding legitimate in most cases, like some sort of desperate, underprepared job interviewee.
As Dr Van der Putten says, ChatGPT is often little more than a “bull*****er on steroids”.
Teaching students about those limitations is the best way to ensure they don’t over rely on it – even in a pinch.